ORIGINAL ARTICLE Open Access

Economic growth, energy consumption,and carbon dioxide emissions in the E7countries: a bootstrap ARDL bound testTeng Tong1, Jaime Ortiz2*, Chuanhua Xu1 and Fangjhy Li1

Abstract

Background: International awareness of the impact of global warming and climate change is increasing.Developing countries face the task of achieving sustainable economic growth while also improving the efficiency oftheir energy consumption. The E7 countries (Brazil, India, Indonesia, Mexico, People’s Republic of China, Russia, andTurkey) are all highly concerned with the promotion of carbon-emission-reduction strategies.

Methods: This research uses a bootstrap autoregressive distributed lag (ARDL) bound test with structural breaks toexamine the cointegration and causality relations between economic growth, energy consumption, and carbondioxide (CO2) emissions in the E7 countries.

Results: There is no cointegration between economic growth, energy consumption, and CO2 emissions for People’sRepublic of China, Indonesia, Mexico, and Turkey. Evidence of cointegration is found for Brazil when CO2 emissions arethe dependent variable and for India and Russia when energy consumption is the dependent variable. For all of the E7countries except Indonesia, short-run Granger causality was found to exist from energy consumption to CO2 emissionsand from economic growth to CO2 emissions for Brazil, India, Mexico, and People’s Republic of China. Short-runGranger causality was also found from economic growth to energy consumption for Brazil, India, Indonesia, Mexico,and People’s Republic of China, and from CO2 emissions to energy consumption for all E7 countries.

Conclusions: The results consistently show that energy consumption is the main cause of CO2 emissions, which hasled to the emergence of global warming problems. Increases in CO2 emissions compel the E7 countries to developsound policies on energy consumption and environmental pollution.

Keywords: Economic growth, Energy consumption, Carbon dioxide emissions, E7 countries, Bootstrap ARDL bound test

JEL classification: C32, Q43, Q53, O44

BackgroundEnergy is at the heart of an economy, because it ultim-ately enables investments and technologies that createjobs and drive growth among countries. The energy cri-ses of 1974 and 1981 made the relationship betweeneconomic growth and energy consumption a focus of

economic research, and this relationship has showngenuine concern for policy analysts [1–3]. Studies of thecausal relationship between the two have reported con-flicting results. For example, Lee [4] used panel datafrom 18 developing countries between 1975 and 2001 toinvestigate the causal relationship between energy con-sumption and gross domestic product (GDP). Grangercausality results showed that energy consumption causedeconomic growth, but not vice versa. Conversely, Jobertand Karanfil [5] tested the relationship between energy

© The Author(s). 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you giveappropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate ifchanges were made. The images or other third party material in this article are included in the article's Creative Commonslicence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commonslicence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtainpermission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to thedata made available in this article, unless otherwise stated in a credit line to the data.

* Correspondence: jortiz22@uh.edu2College of Technology, University of Houston, Isabel C. Cameron Bldg. Suite.227 D, 4235 Cullen Blvd., Houston, TX 77204, USAFull list of author information is available at the end of the article

Energy, Sustainabilityand Society

Tong et al. Energy, Sustainability and Society (2020) 10:20 https://doi.org/10.1186/s13705-020-00253-6

consumption and economic growth in Turkey from1960 to 2003 and found no relationship by Granger orcointegration measures. In contrast, other studies havesuggested a two-way causal relationship between eco-nomic growth and energy consumption [6, 7]. There isobviously a lack of consensus regarding whether the re-lationship between economic growth and energy con-sumption is a one-way causality, a two-way causality, oris completely unrelated, linearly or nonlinearly. Under-standing the relationship between economic growth andenergy consumption is important, however, for helpinggovernments make sound policy decisions.In recent years, international concern about global

warming and climate change has escalated [8–10]. Amajor contributing factor to this problem is the increas-ing level of greenhouse gas carbon dioxide (CO2) world-wide, resulting from the combustion of fossil fuels suchas coal, natural gas, and oil for energy and transportation[11]. According to the 2018 report by British Petroleum(BP), CO2 emissions from energy consumption showedlittle or no growth in 2014–2016 but increased by 1.6%in 2017. It is therefore necessary for all countries to playtheir part in balancing the need to supply the energy theworld needs for growth and prosperity, with the reduc-tion of CO2 emissions. Grippa, Schmittmann, andSuntheim identified that sustainable economic growthreflects the increasing recognition of governments thatclimate change impacts on financial stability, resulting inurgent calls for research into the economic costs of CO2

emissions [12].The 2030 Agenda for Sustainable Development

Goals (SDGs), adopted by the United Nations mem-ber states in 2015, provides a blueprint for inclusiveand sustained economic growth along with sharedprosperity and decent work for people in all coun-tries, taking into account their own set of competen-cies and resource endowments [13]. There exists asense of urgency about the goals of economic growthand sustainable development, and there is greatconsistency and synergy in policy measures. Becausefossil fuels are not a sustainable energy source, eco-nomic growth must be aligned with changes in energysupply and consumption patterns. Within this transi-tion, most countries face the dual task of improvingenergy efficiency while reducing CO2 emissions, butthe options available to developed versus developingcountries are substantially different [14, 15]. Thefocus of developed countries is to improve energy ef-ficiency, whereas developing countries face the greaterchallenge of transitioning their energy structure from theconsumption of coal to the increased use of natural gasand electricity while continuing to rely on fossil fuels.The 2018 Global Energy and CO2 Status Report pre-

pared by the International Energy Agency (IEA) [16],

states that global demand for energy increased by 2.1%in 2017, compared with 1.2% the previous year, and 0.9%on average over the previous 5 years. People’s Republicof China and India accounted for more than 40% of thisgrowth. On the other hand, Price, Waterhouse, and Coo-pers (PWC) [17] have suggested that emerging countrieswill progressively increase their share in the overallGDP. The world economy is projected to grow at an an-nual average rate of 2.6% between 2016 and 2050, andthis growth is expected to be primarily driven by seven-emerging (E7) developing countries: Brazil, India,Indonesia, Mexico, the People’s Republic of China,Russia, and Turkey. The projected annual growth rate ofthese countries is 3.5% over the next 40 years, comparedwith an annual average growth rate of 1.6% for the de-veloped G7 countries: Canada, France, Germany, Italy,Japan, the UK, and the USA.Hawksworth and Cookson [18] coined the term E7 to

describe a group of developing countries with the fastestgrowing populations, which are worldly integrated, withthe goal of becoming economically as strong as the G7countries. By 2014, the E7 countries had already passedthe G7 nations in terms of purchasing power parity [19].Subsequently, PWC predicted that the E7 countries wouldbe 75% larger than the G7 nations in purchasing powerparity terms by 2050. Whether measured by CO2 emis-sions or energy consumption, the E7 countries are playinga larger role in the world energy market (see Fig. 1). In2018, they represented 47% of the world’s population,accounted for 26% of the global GDP, and captured morethan 40% of the global energy consumption. An exampleis the People’s Republic of China, which in 2005 embodied27% of the global demand for energy, a percentage that isexpected to rise to nearly 40% by 2050. In addition,whether measured by total or per-capita share of CO2

emissions, each E7 country ranked among the world’s top20th highest CO2 polluters in 2016.Developing economies, such as the E7, remain particularly

prone to risks resulting from climate change due to theirrapidly increasing levels of energy consumption and the ef-fects of the resulting CO2 pollution [12]. Knowledge of thecausal relationship and the direction between economicgrowth, energy consumption, and CO2 emissions (3E) is ofparamount importance to policymakers. Since E7 countrieshave evolved into influential worldwide economic power-houses, it is necessary to underscore their true 3E relation-ship. The phenomenal economic growth experienced by theE7 countries in the last 20 years has helped reduce the gapbetween them and the G7 countries. In fact, PWC pointedout that the weight of the global economy has systematicallyshifted towards the E7 countries. Coincidentally, the People’sRepublic of China and India stand at the forefront of techno-logical progress and should exert catalytic leading rolesresulting from the growth of their energy sectors [14].

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 2 of 17

This study fills the gap in current literature in severalways. First, it represents the E7 countries covering their1990–2014 period. The main reason for choosing thisspecific sample period is that data before the 90s are notfully reliable across E7 countries. Secondly, it looks forthe longitudinal cointegration and causality relationshipsamong 3E by using the bootstrap autoregressive distrib-uted lag (ARDL) bound test. It therefore adds new em-pirical evidence to the ongoing debate on the 3Erelationship. There are indeed studies on the 3E rela-tionship for some isolated countries within the G20 [21],as well as their nexus for India or Turkey [22, 23]. Thereis, however, limited research on the 3E relationships ofthe E7 countries as a subset. The existing research hasmainly used traditional time series, panel data analysis,or even the standard ARDL model [24, 25]. As men-tioned in Pesaran et al. [26], these methods do not con-sider the two degenerated cases. Hence, our researchuses the bootstrap ARDL bound test to avoid the draw-backs that may provide misleading information for pol-icymakers. At the same time, long-run cointegration andshort-run Granger causality relationships of 3E can beextracted from bootstrap techniques. Our findings con-firm the presence of cointegration between economicgrowth, energy consumption, and CO2 emissions at thetime we also validate the cointegration and causal rela-tionships for all E7 countries.This research makes novel contributions in other as-

pects as well. First, unlike existing cointegration or caus-ality tests, we use the bootstrap ARDL bound method to

test long-term cointegration relationships. It does notneed to distinguish independent variables fromdependent ones, so the method can also be used for atwo-way verification, such as x → y or y → x. Inaddition, the bootstrap ARDL bound test generates crit-ical values to test degradation in the ARDL model.McNown et al. [27] proposed two degenerate cases with-out cointegration relationships, in which only one de-generate case is tested by its critical value. They initiallyproposed this method to not only solve endogeneityproblems, but also find degradation cases, eliminatingthe possibility of uncertain inference.The remainder of the paper is structured as follows.

The “Literature review” section unveils the novelty ofthe topic and presents the literature review. The “Dataand methodology” section reports the data collection,describes the statistics, introduces the traditional ARDLmodel, and details the bootstrap ARDL bound test. The“Results and discussion” section presents the empiricalresults of the unit root tests, the bootstrap ARDL cointe-gration test with structural breaks, and the Grangercausality tests and identifies the economic implicationsand policy recommendations. Finally, the “Conclusions”section reports the main conclusions.

Literature reviewThree general categories of literature focus on the rela-tionships between economic growth, energy consump-tion, and CO2 emissions. The first addresses therelationship between CO2 emissions and economic

0

10

20

30

40

perc

ent o

f wor

ld (

unit:

%)

0

1000

2000

3000

200

400

600

800

Mill

ion

tonn

es o

il eq

uiva

lent

1965 1985 2005 2017

P.R.China

Russia

India

Mexico

Brazil

Indonesia

Turkey

percent

Fig. 1 Energy consumption in E7 countries (1965–2017). Source: BP Statistical Review [20]

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 3 of 17

growth, mainly regarding whether such relationship con-forms to the environmental Kuznets curve (EKC) hy-pothesis. This hypothesis states that there is an invertedU-shaped relationship between certain pollutants andper-capita income [28]. Dinda [29] provides a compre-hensive review of the EKC literature alongside Soytaset al. [30] and Iwata et al. [31] in relation to CO2 emis-sions. Studies on the EKC hypothesis have reached threedifferent conclusions on CO2 emissions. One confirmsthe “inverted U shape theory” [32, 33]. The second, how-ever, postulates a long-term relationship between CO2

emissions and per-capita income and states that it is notan inverted U but an N or other shape [34, 35]. Thethird, known as the non-existence theory, claims thatthere is no relationship between the two [36–38]. Themain problem with these early studies on EKC is thatthey are potentially subject to an omitted variable bias.This occurs when a statistical model leaves out one ormore independent variable(s) that influences thedependent variable and correlates with one or more ofthe included independent variable(s) [39]. Alternatively,other scholars have studied the causality between eco-nomic growth and CO2 emissions. Salahuddin et al. [40]found a two-way Granger causal relationship between thetwo factors, whereas Omri et al. [41] found just a one-wayGranger causality running from CO2 emissions to eco-nomic growth in some European, Central Asian, and LatinAmerican and Caribbean countries. In recent years, differ-ent measurements and econometric methods have drawncontradictory conclusions for various countries.The second category of literature, advanced by Kraft

and Kraft [3], concerns the relationship between energyconsumption and economic growth. Ozturk [42] sug-gests that there are four general types of relationshipstherein. The first, the growth hypothesis, predicts thatenergy consumption will cause economic growth, mean-ing that energy protection policies may stifle GDP [39,43, 44]. The second, the protection hypothesis, claimsenergy consumption has no effect on economic growth,and that energy conservation policies do not necessarilyhave a negative effect on actual GDP [4–46]. The third,the feedback hypothesis, states that energy consumptionand economic growth have a complementary interactionwith each other [47–49]. Finally, the neutral hypothesissuggests that energy consumption does not correlatewith economic growth, meaning that energy conserva-tion policies would have limited influence on economicgrowth [15, 36, 42].The third category of research has studied 3E relation-

ships. Soytas et al. [30] introduced energy consumptioninto the relationship between GDP and CO2 emissionsto circumvent a potential omitted variable bias. Theseresults show that, in the USA, income does not causeCO2 emissions, but energy consumption does. Apergis

and Payne [6] found that for six Central American coun-tries over the period 1971–2004, long-term equilibriumin energy consumption had a positive and statisticallysignificant effect on CO2 emissions, while real GDP ex-hibited an inverted U-shape associated with the EKC hy-pothesis. Pao and Tsai [50] studied Brazil, Russia, India,and the People’s Republic of China (BRIC) from 1971 to2005. Their findings suggest that, except for Russia in1990–2005, there was a bidirectional strong causality be-tween energy consumption and CO2 emissions, and a bi-directional mild long-run causality between energyconsumption and GDP. They also found a strong short-run unidirectional causality between CO2 emissions andenergy consumption as well as GDP.Extant studies have indeed surveyed much of the 3E

relationships through various econometric techniquesacross continents and regions. Table 1 summarizes theresults found for several countries since 2010. It be-comes clear that research on the 3E relationships hasbeen widely scrutinized via methodologies includingpanel cointegration tests [8, 10, 15, 52, 53] and nonlinearcausality tests [51]. In terms of surveying cointegrationand causality between macroeconomic variables forsmall samples, panel data analysis seems to be themethod of choice.Another econometric approach is the ARDL bound

test proposed by Pesaran et al. [26]. Using this method,Jalil and Mahmud [55] investigated the causality betweenCO2 emissions and economic growth using time seriesdata pertaining to 1975–2005 for the People’s Republicof China. The results show that a one-way causality runsfrom economic growth to CO2 emissions. Ozturk andAcaravci [23] examined the long-run and causal rela-tionship of 3E in Turkey over the period 1968–2005.Their results show the existence of a long-run relation-ship between the variables, but neither CO2 emissionsnor energy consumption cause GDP. Asongu et al. [24]revisited the 3E relationship in 24 African countriesusing a panel ARDL approach. They found the presenceof a long-run relationship among 3E where causalityruns from GDP and CO2 emissions to energy consump-tion. However, there were no short-run causalities fromCO2 emissions to GDP, from GDP to energy consump-tion, or from energy consumption to GDP. Magazzino[56] looked at the relationship among GDP, CO2 emis-sions, and energy use in South Caucasus countries andTurkey over the 1992–2013 period using a time seriesARDL approach and uncovered a long-run relationshipbetween these variables. Alam et al. [57] examined theimpacts of income, energy consumption, and populationgrowth on CO2 emissions, using time series data from1970 to 2012, for Brazil, India, Indonesia, and the Peo-ple’s Republic of China. Their results showed that bothGDP and energy consumption have a positive effect on

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 4 of 17

CO2 emissions in these four countries. McNown et al.[27] pointed out that the ARDL bound test has the ad-vantage of solving endogeneity problems and eliminatingthe possibility of inconclusive inference.Hereafter, an increasing number of studies began using

the bootstrap technique and causality based on theARDL model. Examples include Goh et al.’s study on theenergy consumption-economic growth relationship for22 OECD countries [58], and Lin et al.’s study on thecoal consumption, CO2 emissions, and economic growthnexus for the People’s Republic of China and India [59].However, despite the abovementioned advantages, no re-search using the bootstrap ARDL bound test has beencarried out on the E7 countries. Most studies on the 3Erelationship identify a positive cause-and-effect relation-ship between CO2 and energy consumption, especiallythose focusing on developing countries [8, 10, 24, 45, 52,55]. Unfortunately, the results lack sufficient strength forverification of a linear or nonlinear causality between

CO2 emissions and economic growth or between energyconsumption and economic growth, in order to confirmthe EKC hypothesis.

Data and methodologyThe three variables used in this study are per-capita GDPin constant 2010 US$, per-capita total primary energyconsumption in a kilogram of oil equivalent (kgoe), andper-capita CO2 emissions in metric tons (mt). These vari-ables are examined for the E7 countries over the period1971 to 2014, except for Russia (1992–2014) because ofdata irregularities. Data for all sampled countries were ob-tained from the World Development Indicators1 of theWorld Bank. Logarithms were applied to all variables toameliorate heteroscedasticity issues. The inter-quartile

Table 1 The 3E nexus found for various countries, 2010–2018Authors Variables Region Period Methodology Findings

Pao and Tsai [50] CO2, EC, GDP BRIC countries 1971–2005 VECM; VAR - Bidirectional strong Granger causality between EC and CO2

output- Bidirectional mild long-run relationship between ECand GDP

- Causality running from EC to GDP in the short-run- Strong negative causality running from CO2 outputto GDP

Wang et al. [10] CO2, EC, GDP 28 Provinces in thePeople’s Republicof China

1995–2007 Panel cointegration;panel VECM

- Bidirectional causality between CO2 and EC, alsobetween EC and GDP

- EC and economic growth are the long-run causesfor CO2 emissions

- CO2 emissions and economic growth are thelong-run causes for EC

Wang et al. [51] GDP, EC, CO2 People’s Republic ofChina

1978–2012 Linear and nonlinearcausality

- Linear and nonlinear unidirectional causality from CO2

emissions to GDP- Linear and nonlinear bidirectional causality betweenEC and CO2 emissions

- Unidirectional linear causality from EC to GDP- Nonlinear unidirectional causality from GDP to EC

Magazzino [52] GDP, EC, CO2 South Caucasusand Turkey

1992–2013 Panel VAR - CO2 has a negative effect on energy use- Real GDP has no effect on energy use- Energy use has no effect on real GDP

Chen et al. [8] CO2, EC, GDP 188 countries 1993–2010 Panel cointegration;VECM

- Long-run relationships between GDP, EC, and CO2 for allcountries

- EC negatively affects GDP in the world as a whole anddeveloping countries, but not in developed countries

- Unidirectional causality from EC to CO2 exists both ondeveloping and developed countries

Magazzino [53] GDP, EC, CO2 19 APEC countries 1960–2013 Panel VAR - No long-run relationship between GDP, EC and CO2

- No causal relationship between real GDP and energyuse

- Causal relationship exist between GDP and energyin 9 countries

Mirza and Kanwal [54] GDP, EC, CO2 Pakistan 1971–2009 ARDL; VECM - Short-run, long-run, and strong Granger causalityresults indicate bidirectional causalities among EC,GDP, and CO2 emissions

Doğan [15] CO2, FD, EC, GDP E7 countries 1990–2014 Panel cointegration - No long-term relationship between CO2 emissionsand FD

- Positive effect of EC on CO2 emissions- Positive effect of economic growth on CO2 emissionsin long-term

Pao and Chen [21] CO2 EC, GDP G20 1991–2016 Panel cointegration - Long-run equilibrium relationship among CO2 emissions,consumption of fossil fuels, GDP, and clean energyconsumption

Note: ARDL autoregressive distributed lag procedure, EC energy consumption, FD financial development, VAR Vector Autoregressive Model, VECM Vector Error Correction Model

1World Bank Indicators (WBI), see more details: https://databank.worldbank.org/source/world-development-indicators

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 5 of 17

range (IQR) shows the absence of outliers for the variables(see Table 2).We used the bootstrap ARDL model to examine the

nexus among economic growth, energy consumption,and CO2 emissions for the E7 countries. The bootstrapARDL bound test with structural breaks uses theprinciple of self-regression and multiple loop calibrationsto approximate the time series data for verification ofthe expected results.

Unit root testTime series analysis requires firstly performing the unitroot test to check whether the collected data is station-ary, in order to avoid false regressions. A steady statemeans that statistics such as the mean and variance donot change over time. In other words, the self-covariance and the variance are fixed finite constantvalues. We applied the augmented Dickey–Fuller (ADF),Phillips and Perron (PP) [60], Elliott et al, (DF-GLS)[61], and the Kwiatkowski–Phillips–Schmidt–Shin(KPSS) [62] unit root tests for LGDP, LEC, and LCO2.In addition, we also used the Zivot and Andrew (Z-A)unit root test [63] which allows for breaks at a point inthe intercept term and/or the line trend items [63], aswell as the Lee and Strazicich (LS) [64] unit root testwith two structural breaks to strengthen the validity ofthe univariate unit root tests.

Bootstrap ARDL bound testThe ARDL bound test allows a better understandingof the cointegration of the series in the model. AMonte Carlo simulation was used to estimate the sizeand power characteristics of endogeneity. The asymp-totic threshold of the simulation has only a small ef-fect. If the re-sampling process is applied properly,the pilot-to-test ratio is determined, and the asymp-totic check in the ARDL bound test based on the sizeand power characteristics performs better. Theasymptotic threshold describes the extension of thevalidation framework under an alternative degrad-ation, as well as the threshold itself generated by thebootstrap ARDL bound test that is based on theGranger causality test. The standard Granger causalitydetermines the direction of the short-term causal re-lationship. If y is due to a variable, no agreement isfound between y and x. The Granger causality test ofx → y should only include the hysteresis difference ofx. If there is a cointegration relationship between thevariables, then the dependent and independent vari-ables form a fixed linear combination. The hysteresisterm can be considered as I (0), and the Grangercausality test of x → y should include the hysteresisdifference of x and the hysteresis level of x. The

ARDL bound test [26] has a time series of mixed in-tegration sequences that is defined as:

yt ¼ aþXk

i¼1

αiyt−i þXk

i¼1

βixt−i þXl

j¼1

ψ jDt; j þ μt ð1Þ

In the case of an exogenous weak regression, these re-gression factors are not affected by the variables in thelong-run. The model neither precludes the existence ofcointegration between regressions nor assumes that thedependent variable of the regression exhibits short-termGranger causality.The ARDL model with three variables is:

yt ¼ aþXk

i¼1

αiyt−i þXk

i¼1

βixt−i þXk

i¼1

δizt−i

þXl

j¼1

ψ jDt; j þ μt ð2Þ

where i and j are the indicators of the lag period, i = 1,2,…, k; j = 1, 2,…, k. t represents time t = 1, 2,…, T. Theyt in the equation is the interpreted variable and xt is theexplanatory variable; Dt,j is a dummy variable. The errorterm is μt, and Eq. (2) is rewritten and expanded in anerror correction representation as:

Δyt ¼ γ0 þXk−1

i¼1

γ1Δyt−i þXk−1

i¼1

γ2Δxt−i

þXk−1

i¼1

γ3Δzt−i þXl

j¼1

γ4Dt; j þ θ1yt−1

þ θ2xt−1 þ θ3zt−1 ð3Þ

where θ1 ¼ −ð1−Pki¼0αiÞ; θ2 ¼

Pki¼0βi; θ3 ¼

Pki¼0δi;

other parameters are the function values of the original

parameters in Eq. (2).According to Pesaran et al. [26], the cointegration test

needs to be an F test or a t test. The following assump-tions are made:H0: θ1 = θ2 = θ3 = 0 or H0:θ1 = 0McNown et al. [27] proposed adding the original

ARDL model to a lag period for independent variableswhere the null hypothesis is θ2 = θ3 = 0. Doing this cre-ates more complete conditions for testing the cointegra-tion relationship proposed by Pesaran et al. [6]. Then,the following three null hypotheses must be rejected:

The null hypothesis error term F1 is tested as H0: θ1 =θ2 = θ3 = 0.The t test for the lag dependent variable is H0: θ1 = 0.The F2 test for the lag independent variable is H0: θ2 =θ3 = 0.

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 6 of 17

Table

2Summarystatistics

Econ

omic

variables

Brazil

India

Indo

nesia

Mexico

Peop

le’sRepu

blicof

China

Russia

Turkey

LGDP

LEC

LCO2

LGDP

LEC

LCO2

LGDP

LEC

LCO2

LGDP

LEC

LCO2

LGDP

LEC

LCO2

LGDP

LEC

LCO2

LGDP

LEC

LCO2

Mean

9.039

6.926

0.482

6.442

5.923

−0.298

7.484

6.295

−0.039

8.974

7.217

1.301

6.922

6.778

0.933

9.005

8.429

2.438

8.800

6.806

0.864

Med

ian

9.032

6.898

0.430

6.321

5.900

−0.252

7.562

6.385

0.058

8.980

7.286

1.340

6.849

6.656

0.865

8.958

8.419

2.421

8.788

6.843

0.981

Max.

9.386

7.303

0.953

7.406

6.457

0.548

8.214

6.784

0.940

9.162

7.414

1.471

8.717

7.713

2.023

9.376

8.585

2.638

9.496

7.369

1.502

Min.

8.539

6.571

0.049

5.863

5.591

−1.012

6.690

5.694

−1.107

8.612

6.685

0.855

5.471

6.142

0.041

8.614

8.289

2.315

8.177

6.100

−0.123

Std.

dev.

0.185

0.173

0.201

0.474

0.255

0.465

0.429

0.371

0.531

0.144

0.183

0.161

1.038

0.466

0.589

0.272

0.083

0.081

0.348

0.358

0.444

Skew

ness

−0.194

0.386

0.342

0.545

0.471

0.064

−0.168

−0.223

−0.207

−0.750

−1.533

−1.556

0.171

0.704

0.426

0.019

0.165

0.660

0.216

−0.224

−0.461

Kurtosis

3.469

2.720

2.726

2.065

2.184

1.825

1.975

1.511

2.195

2.805

4.294

4.472

1.749

2.349

2.177

1.477

1.946

2.937

2.087

2.068

2.246

IQR

0.166

0.201

0.277

0.769

0.378

0.749

0.626

0.690

0.791

0.193

0.105

0.103

1.843

0.656

0.886

0.484

0.163

0.114

0.478

0.463

0.576

Note:

LGDP,

LEC,and

LCO2represen

tpe

r-capita

real

GDP,

per-capita

energy

consum

ption,

andpe

r-capita

CO2em

ission

sallinloga

rithm

form

,respe

ctively.Th

ede

scrip

tivestatisticsareba

sedon

thedifferen

cesin

the

loga

rithm

sof

each

varia

ble

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 7 of 17

Based on three null hypotheses, McNown et al. [27]explain two degenerates of Pesaran et al. [26]. The twodegeneration cases are as follows:

Degenerate case #1. The F1 test and the t test forthe lag dependent variable are significant, but the F2test for the lag independent variable is notsignificant.Degenerate case #2. The F1 and F2 tests for the lagdependent variable are significant, but the t test for thelag dependent variable is not significant.

Pesaran et al. [26] excluded degeneration case #1, andif they did not consider the integration order of thedependent variable, it must be I (1). To solve this prob-lem, McNown et al. [27] used the bootstrap ARDLbound test as an additional test of the lagging independ-ent coefficient. If there is a cointegration relationship be-tween the dependent variable and an independentvariable, the above three virtual hypotheses will simul-taneously be rejected, and the interpreted and explana-tory variables will be stable linear coincidences.If, after testing the long-term 3E relationships, we find

no cointegration relationships between y, x, and z, weuse the Granger causality test for x and z, which shouldinclude the difference in hysteresis on x or z. We test γ2= 0 or γ3 = 0 in Eq. (3). However, if there is cointegra-tion between the dependent variable and the independ-ent variable, this means that they form a fixed linearcombination. In this case, the short-term relationshiptest should include the hysteresis difference of x or zand the hysteresis level of x or z; that is, test γ2 and θ2orγ3 and θ3.

Results and discussionUnit root testAs mentioned, a prerequisite of the bootstrap ARDLbound test is the performance of unit root tests toverify the three variables are stationary. We used theADF, DF-GLS, PP, and KPSS tests to achieve such atask. Table 3 reveals that all variable sequences wereno more integrated than I (1) with the ADF, DF-GLS,and PP tests. However, the KPSS test for LGDP forthe People’s Republic of China and the LEC andLCO2 for Russia does reject the null hypothesis ofstationarity. Similarly, Tables 4 and 5 present the re-sults of the Z-A and LS unit root tests. Table 4 re-ports the stationarity of variables employed in leveland first differences where a break at a point in theintercept and/or the slope of the trend function is allowed.All variables are I (1) process or less. Table 5 shows thevariables rejecting the null hypothesis that the series has aunit root on first difference. We therefore conclude they

are all stationary at either I (0) or I (1), which is consistentwith the premise of bootstrap ARDL bound test.2

Bootstrap ARDL cointegration test with structural breaksThe bootstrap ARDL bound test provides a betterinsight into the cointegration status of the series whenstructural breaks are identified [27]. We followed Baiand Perron [65] to determine the structural breaks inthe equation for each country. Multiple breakpointswere plugged into the ARDL model as dummy variables.Table 6 reports the estimation and testing of Eq. (3)using the bootstrap ARDL bound test. Each ARDL equa-tion passed all diagnostic tests for autocorrelation, non-normality, and heteroscedasticity. These lag lengths weredetermined using the AIC. The F1

*, F2*, and t* columns

present values for the 0.10 significance level, generatedby the ARDL bootstrap procedure. The bootstrap ARDLbound test allows for the renormalization of the ARDLequation to treat every series as a dependent variable. Adegenerate case #2 exists for Indonesia and the People’sRepublic of China where energy consumption serves asthe dependent variable, and Mexico where GDP servesas the dependent variables, since the F1 and F2 tests forthe lagged dependent variable were significant but theirt tests were not. For the People’s Republic of China,Indonesia, Mexico, and Turkey, no cointegration wasfound for economic growth, energy consumption, orCO2 emissions. Evidence of cointegration was found forBrazil when CO2 emissions are the dependent variableand for India and Russia when energy consumption wasthe dependent variable.

Granger causality testsTable 7 shows the long-term causality tests on Brazil,India, and Russia based on the cointegration analysis.The results show that economic growth has a positiveGranger causality effect on energy consumption onlyfor India, implying that its economic growth willlargely be accompanied by increasing energy con-sumption in the long-run. Table 8 and Fig. 2 presentthe short-run Granger causality results between eco-nomic growth, energy consumption, and CO2 emis-sion for the E7 countries. There is evidence of a firmshort-run Granger causality from energy consumptionto economic growth for Brazil, India, Indonesia,Mexico, and Turkey. Short-run Granger causalityfrom CO2 emissions to economic growth is presentfor all E7 countries, suggesting that even though nocointegration exists when economic growth is the

2We performed multiple linear and nonlinear unit root tests toimprove the alleged deficiency of the univariate unit root tests and toidentify unknown structural breaks. These results available uponrequest were all consistent.

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 8 of 17

Table

3Univariate

unitroot

tests

Cou

ntries

Variables

Atlevel

Atfirstdifference

ADF

DF-GLS

PPKPSS

ADF

DF-GLS

PPKPSS

Brazil

LGDP

−3.3939

(0)*

−1.8728

(0)

−3.3855

(3)*

0.8022

(3)**

*−4.7558

(0)**

*−2.2785

(1)**

−4.7003

(2)**

*0.2139

(0)

LEC

−1.6069

(0)

−1.5446

(0)

−2.0561

(3)

0.1540

(3)**

−5.5424

(0)**

*−5.4056

(0)**

*−5.5424

(0)**

*0.1317

(2)

LCO2

−2.1627

(2)

−2.5969

(2)

−2.2599

(3)

0.1193

(3)*

−4.7184

(0)**

*−2.6677

(1)**

*−4.7184

(0)**

*0.0974

(3)

India

LGDP

−1.3653

(0)

−0.5646

(0)

−1.3299

(3)

0.2185

(2)**

*−7.9501

(0)**

*−3.7990

(0)**

*−7.9501

(0)**

*0.0410

(3)

LEC

−0.2063

(0)

−1.3863

(3)

−0.3839

(3)

0.1728

(3)**

−6.1489

(0)**

*−4.2217

(0)**

*−6.2067

(3)**

*0.0811

(3)

LCO2

−1.8144

(0)

−1.7373

(0)

−2.0078

(3)

0.8395

(5)**

*−6.3012

(0)**

*−6.1957

(0)**

*−6.3295

(3)**

*0.2011

(0)

Indo

nesia

LGDP

−2.5628

(1)

−2.3342

(1)

−2.1648

(1)

1.1831

(3)**

*−4.8001

(0)**

*−4.8759

(0)**

*−4.7678

(2)**

*0.1354

(1)

LEC

−1.2409

(0)

−1.4495

(0)

−1.2409

(0)

1.1781

(3)**

*−6.6338

(0)**

*−6.5387

(0)**

*−6.6734

(4)**

*0.1645

(0)

LCO2

−3.7588

(1)**

−2.3167

(2)

−2.9585

(3)

1.4971

(2)**

*−6.0189

(1)**

*−5.9762

(1)**

*−6.0182

(3)**

*0.2978

(3)

Mexico

LGDP

−2.8202

(0)

−1.9435

(0)

−2.8974

(2)

0.7749

(2)**

*−5.3440

(0)**

*−4.6491

(0)**

*−5.3012

(3)**

*0.2574

(0)

LEC

−2.6150

(0)

−1.1455

(0)

−2.5758

(1)

0.6575

(3)**

−5.1826

(0)**

*−5.1506

(0)**

*−5.1826

(0)**

*0.1108

(1)

LCO2

−2.2049

(0)

−1.4732

(0)

−2.1412

(1)

0.5729

(3)**

−7.8209

(0)**

*−7.0476

(0)**

*−7.7964

(1)**

*0.0709

(0)

Peop

le’sRepu

blicof

China

LGDP

−3.1892

(1)

−1.5464

(1)

−3.6616

(3)*

0.1815

(2)**

−4.3604

(0)**

*−4.1830

(0)**

*−4.3309

(3)**

*0.1215

(2)*

LEC

−1.3500

(1)

−1.4611

(1)

−0.8771

(3)

0.1855

(3)**

−3.7419

(0)**

−3.8233

(0)**

*−3.7255

(1)**

0.0819

(3)

LCO2

−2.2675

(1)

−2.3137

(1)

−1.5660

(3)

0.1484

(2)**

−3.4145

(0)**

−3.4539

(0)**

*−3.4620

(1)**

0.1172

(0)

Russia

LGDP

−3.1829

(1)

−2.1944

(1)

−3.4788

(2)

0.5752

(3)**

−2.4446

(0)**

−2.2988

(0)**

−2.3641

(1)**

0.3231

(2)

LEC

−4.0600

(0)**

−2.5831

(0)

−4.7772

(2)

0.1674

(2)**

−3.3568

(0)*

−2.9837

(0)**

*−3.3568

(2)*

0.3821

(2)*

LCO2

−3.9379

(0)**

−2.5851

(0)

−3.9379

(2)**

0.1689

(0)**

−3.3132

(0)*

−2.9842

(0)**

*−3.3132

(0)*

0.3598

(2)*

Turkey

LGDP

−2.2229

(0)

−2.3089

(0)

−2.3966

(1)

0.9429

(2)***

−4.8594

(3)**

*−5.0686

(0)**

*−6.8309

(3)**

*0.0866

(2)

LEC

−2.9710

(0)

−2.4336

(0)

−3.0543

(1)

0.9327

(3)**

*−6.6989

(0)**

*−5.5857

(0)**

*−6.7042

(3)**

*0.1314

(3)

LCO2

−2.9952

(0)

−1.7598

(0)

−2.9864

(3)

0.9260

(3)**

*−6.6613

(0)**

*−4.3947

(0)**

*−6.6666

(3)**

*0.1299

(5)

Note:

Asterisks

*** ,

**an

d*indicate

the1%

,5%

and10

%levels,respe

ctively

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 9 of 17

Table

4ZivotandAnd

rew

unitroot

tests

Cou

ntries

Variables

Atlevel

Atfirstdifference

Intercep

t(lag)

breakyear

Tren

dandIntercep

t(lag)

breakyear

Intercep

t(lag)

breakyear

Tren

dandintercep

t(lag)

breakyear

LGDP

−4.595(1)2007

*−4.671(1)1998

−6.047(0)1981

***

−6.638(0)1984

***

Brazil

LEC

−3.244(2)2007

−4.009(2)2000

−6.537(0)1980

***

−6.778(0)1983

***

LCO2

−3.756(2)1981

−3.626(2)1980

−5.514(0)1980

***

−6.177(0)1984

***

LGDP

−2.818(0)1979

−3.133(0)1993

−8.054(0)1979

***

−8.274(0)2004

***

India

LEC

−2.931(0)2007

−4.066(0)2001

−7.162(0)2004

***

−7.213(0)2006

***

LCO2

−2.505(0)1998

−3.532(0)2001

−7.014(0)1997

***

−7.084(0)1997

***

LGDP

−8.807(1)1998

***

−7.589(1)1998

***

−5.494(0)1997

***

−6.512(0)1998

***

Indo

nesia

LEC

−5.034(0)1990

**−5.980(0)1990

***

−7.357(0)1986

***

−7.416(0)1990

***

LCO2

−4.362(1)1998

−4.562(1)1982

−6.364(1)1980

***

−6.524(1)1979

***

LGDP

−3.365(1)1979

−4.120(1)1983

−6.579(0)1982

***

−6.687(0)1982

***

Mexico

LEC

−4.261(0)1978

−4.551(0)1978

−6.538(0)1982

***

−6.642(0)1982

***

LCO2

−3.722(0)1978

−4.637(0)1983

−9.563(0)1983

***

−9.456(0)1983

***

LGDP

−3.666(1)2005

−3.374(1)1979

−5.008(0)1978

**−5.111(0)1986

**

Peop

le’sRepu

blicof

China

LEC

−4.574(1)2003

−4.141(1)1997

−4.699(0)2002

−5.783(0)2003

***

LCO2

−4.114(1)2003

−4.651(1)1997

−4.454(0)2002

−5.235(0)2002

**

Russia

LGDP

−4.536(1)2009

−4.936(1)2009*

−4.216(0)1999

−4.910(0)2000

*

LEC

−4.178(0)2003

−3.800(0)1997

−4.666(0)1996

***

−5.706(0)1999

***

LCO2

−4.059(0)2003

−3.680(0)1997

−4.304(0)1996

−4.987(0)1999

*

LGDP

−3.209(0)1979

−3.459(0)2001

−6.510(0)2003

***

−6.576(0)1981

***

Turkey

LEC

−4.365(0)1979

−4.210(0)1986

−6.799(0)1978

***

−6.940(0)1982

***

LCO2

−4.343(0)1985

−4.380(0)1985

−6.679(0)1978

***

−6.998(0)1981

***

Note:

Asterisks

***,**,*

indicate

the1%

,5%

and10

%sign

ificancelevels,respe

ctively.Th

enu

mbe

rsin

parenthe

sesrepresen

ttheop

timal

lagleng

thforZ-Atest

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 10 of 17

dependent variable, CO2 emissions and energy con-sumption are still important short-run determinantsof economic growth for them.Similarly, short-run Granger causality exists from en-

ergy consumption to CO2 emissions across all E7 coun-tries except Indonesia, and from economic growth toCO2 emissions for Brazil, India, Mexico, and the People’sRepublic of China. Regarding energy consumption,Granger causality is observed from economic growth toenergy consumption for Brazil, India, Indonesia, Mexico,and the People’s Republic of China in the short-run, andfrom CO2 emissions to energy consumption for all E7countries.

Economic implicationsThere is cointegration between energy consumption,economic growth, and CO2 emissions for Brazil, India,and Russia, but not for Indonesia, Mexico, the People’sRepublic of China, and Turkey. Evidence of cointegra-tion is found for Brazil when CO2 emissions are thedependent variable, which suggests that economicgrowth and energy consumption are the main factors ofincreased CO2 emissions. As for India and Russia, coin-tegration exists when energy consumption is the

dependent variable, which suggests that economicgrowth is the main determinant of increased energy con-sumption in the long-run. However, no cointegration isfound when economic growth serves as the dependentvariable, which means neither energy consumption norCO2 emissions can account for the economic growth forthe E7 countries in the long-run. The Granger causalitybased on the bootstrap ARDL bound test confirms a bi-directional causality relationship between energy con-sumption and CO2 emissions for all E7 countries exceptIndonesia, which is consistent with the fact that energyconsumption does produce CO2 emissions.The economic implications of the above results are

that energy consumption, as the main cause of CO2

emissions, has led to the emergence of global warmingand the need for developing countries to use energymore efficiently. These countries should follow energyconservation and energy efficient policies to control CO2

emissions. As representatives of developing countries, anincrease in CO2 emissions compels E7 countries to de-velop stringent policies on energy consumption. Inaddition, they should consider implementing CO2 tax-ation strategies, transforming to renewable energy, andpromoting the use of new sources of energy. These

Table 5 Lee and Strazicich LM unit root tests results on first difference

Countries Variables Intercept Trend and Intercept

Test statistic (lag) TB1 TB2 Test statistic (lag) TB1 TB2

Brazil LGDP − 3.418 (0)* 1977 1979 − 7.362 (4)*** 1978 1981

LEC − 5.160 (4)*** 1979 2006 − 7.045 (4)*** 1979 1987

LCO2 − 5.491 (0)*** 1978 1982 − 5.883 (0)*** 1979 1982

India LGDP − 4.616 (1)*** 1983 1987 − 8.031 (0)*** 1977 1980

LEC − 4.737 (3)*** 1984 1994 − 6.374 (3)** 1997 2007

LCO2 − 4.721 (1)*** 1996 2000 − 6.797 (0)** 1976 1990

Indonesia LGDP − 5.646 (0)*** 1979 1995 − 9.588 (1)*** 1995 1999

LEC − 6.212 (0)*** 1977 1982 − 6.745 (1)** 1988 1994

LCO2 − 6.105 (1)*** 1986 1999 − 6.862 (1)*** 1986 2008

Mexico LGDP − 6.483 (0)*** 1977 1981 − 7.304 (4)*** 1979 1987

LEC − 5.077 (0)*** 1977 1979 − 6.545 (4)** 1980 2000

LCO2 − 7.969 (0)*** 1977 1980 − 9.657 (0)*** 1977 1982

People’s Republic of China LGDP − 4.431 (2)*** 1976 1988 − 5.983 (5)* 1986 2003

LEC − 4.151 (0)*** 1977 2009 − 7.656 (3)*** 1988 2000

LCO2 − 3.851 (3)** 1993 1997 − 6.330 (3)* 1993 2000

Russia LGDP − 3.806 (0)** 1998 2007 − 11.61 (5)*** 2003 2007

LEC − 4.033 (4)** 2001 2007 − 6.262 (2)** 1997 2007

LCO2 − 4.901 (2)*** 1997 2009 − 6.772 (2)** 1997 2007

Turkey LGDP − 5.423 (0)*** 1977 1979 − 7.521 (4)*** 1992 1998

LEC − 6.549 (0)*** 1977 2005 − 6.887 (0)*** 1977 1987

LCO2 − 6.576 (0)*** 1978 2003 − 6.516 (0)** 1977 1985

Note: Asterisks ***, ** and * indicate the 1%, 5% and 10% significance levels, respectively. Numbers in parentheses represent the optimal number of lag terms,break date, test statistics, and critical t values. TB1 and TB2 represent time break 1 and time break 2

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 11 of 17

Table

6Results

forbo

otstrapARD

Lcointegrationtest

Cou

ntry

Dep

ende

ntvariable|inde

pend

ent

variable

F 1F 1

*t

t*F 2

F 2*

Dum

myvariables

Result

Brazil

(LGDP|LEC|LCO2)

3.087

3.178

−2.798

−1.585

4.459

3.287

D77,D

04,

Nocointegration

(LEC

|LCO2|LG

DP)

1.451

2.747

0.810

−1.456

0.848

2.788

D77,D

97,D

07Nocointegration

(LCO2|LG

DP|LEC)

5.738

4.087

−3.132

−2.786

8.599

4.913

D77,D

96,D

09Cointeg

ratio

n

India

(LGDP|LEC|LCO2)

3.445

2.760

−0.097

−1.188

0.240

2.964

D88,D

98,D

07Nocointegration

(LEC

|LCO2|LG

DP)

7.820

4.643

−3.103

−1.152

8.944

5.918

D85,D

95,D

02,D

08Cointeg

ratio

n

(LCO2|LG

DP|LEC)

0.467

4.194

−0.399

−2.917

0.314

5.435

D83,D

90,D

96,D

07Nocointegration

Indo

nesia

(LGDP|LEC|LCO2)

0.575

3.210

−1.179

−2.267

0.688

3.752

D79,D

90,D

06Nocointegration

(LEC

|LCO2|LG

DP)

3.238

3.004

−0.444

−1.925

3.956

3.272

D78,D

90,D

00Deg

eneratecase

#2

(LCO2|LG

DP|LEC)

0.775

3.643

−1.065

−2.310

1.151

4.476

D77,D

91,D

01,D

07Nocointegration

Mexico

(LGDP|LEC|LCO2)

4.309

2.848

−0.114

−2.046

4.458

3.071

D79,D

98Deg

eneratecase

#2

(LEC

|LCO2|LG

DP)

4.082

4.858

−0.561

−3.306

0.680

6.421

D79,D

99Nocointegration

(LCO2|LG

DP|LEC)

3.201

4.625

−2.422

−0.893

2.260

2.798

D78,D

01Nocointegration

Peop

le’sRepu

blicof

China

(LGDP|LEC|LCO2)

0.934

3.371

1.183

−2.306

0.840

3.800

D84,D

94,D

05Nocointegration

(LEC

|LCO2|LG

DP)

5.155

2.965

1.351

−1.923

7.022

3.244

D77,D

87,D

94,D

03,D

09Deg

eneratecase

#2

(LCO2|LG

DP|LEC)

2.973

3.173

−2.828

−2.248

4.451

3.669

D77,D

85,D

93,D

03,D

09Nocointegration

Russia

(LGDP|LEC|LCO2)

1.744

13.258

−1.003

−2.609

1.687

5.368

D04,D

07Nocointegration

(LEC

|LCO2|LG

DP)

14.826

3.283

1.935

−1.671

17.601

3.131

D95,D

06Cointeg

ratio

n

(LCO2|LG

DP|LEC)

9.542

42.776

−2.579

−1.326

8.919

21.066

D95,D

06Nocointegration

Turkey

(LGDP|LEC|LCO2)

1.537

3.414

1.636

−2.336

1.129

4.108

D75,D

90,D

05Nocointegration

(LEC

|LCO2|LG

DP)

0.939

3.657

−1.555

−2.498

1.289

4.592

D73,D

87,D

96,D

06Nocointegration

(LCO2|LG

DP|LEC)

1.439

3.950

1.487

−0.372

2.024

4.933

D72,D

86,D

95,D

05Nocointegration

Note:

F 1istheFstatistic

forthecoefficientsof

y(−

1),x

1(−

1),and

x 2(−

1).F

2istheFstatistic

forthecoefficientsof

x 1(−

1)an

dx 2

(−1);t

deno

testhetstatistic

forthecoefficient

ofy(−

1).D

##refers

tothedu

mmy

forthat

year

*Significan

ceat

the.10levelb

ased

oncriticalv

alue

sge

neratedfrom

theARD

Lbo

otstrapbo

undtest

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 12 of 17

countries can also address global warming and climaticinstability by setting realistic mitigation commitments ofCO2 emissions.The Granger causality test supports, for Brazil and

Mexico, the presence of a positive bidirectional causalrelationship between energy consumption and CO2

emissions, and a negative bidirectional causal relation-ship between CO2 emissions and economic growth.These results support the conventional belief that energyconsumption and economic growth closely interact witheach other and confirm the fact that Brazil and Mexicoare energy-dependent economies. These countries re-quire supply-side energy infrastructure policies and en-ergy conservation policies to prevent furtherenvironmental degradation. The results also reveal a bi-directional causal relationship between energy consump-tion and CO2 emissions.India portraits bidirectional causal relationships be-

tween energy consumption and economic growth, eco-nomic growth and CO2 emissions, and CO2 emissionsand energy consumption. However, our results do notsupport the presence of reverse relationships. As in thecase of Brazil and Mexico, India is an energy-dependenteconomy that faces the complex task of conserving en-ergy to reduce CO2 emissions while stimulating eco-nomic growth.Indonesia depicts a negative bidirectional causality be-

tween energy consumption and economic growth. Itscurrent shortage of energy consumption for construc-tion, manufacturing, and transportation is restraining fu-ture economic growth. The unidirectional relationshiprunning from CO2 emissions to energy consumptionand economic growth compels the enactment of conser-vation policies to reduce end energy consumption.The People’s Republic of China shows bidirectional

causalities between energy consumption and CO2 emis-sions, and CO2 emissions and economic growth alongwith a unidirectional causality from economic growth toenergy consumption. There are bidirectional causalitiesbetween economic growth and CO2 emissions. Empiricalevidence reveals that environmental degradation has acausal impact on economic growth, while a persistentdecline in environmental quality explains the severenegative externalities that surround declines in product-ivity. It is imperative for the People’s Republic of China

to take urgent action to decrease CO2 emissions by re-ducing its demand for energy. Similarly, Russia exempli-fies a positive bidirectional Granger causality betweenCO2 emissions and energy consumption as well as anegative unidirectional Granger causality from CO2

emissions to economic growth. Turkey presents a posi-tive bidirectional Granger causality of CO2 emissionsand energy consumption like the People’s Republic ofChina and Russia, while there is a unidirectional Grangercausality from energy consumption and CO2 emissionsto economic growth. These energy-dependent econ-omies need supply-side energy infrastructure policiesand energy conservation policies compatible with a sus-tainable economic growth.This research found an inextricable unidirectional or

bidirectional short-run causality between economicgrowth and CO2 emissions for all E7 countries. Brazil,India, and Russia showed long-run relationships betweenenergy consumption and economic growth in respect ofCO2 emissions, when the latter is the dependent vari-able. In Brazil, India, and Mexico, CO2 emissions woulddecrease over the time when GDP increases. Thus, itcan be argued that these three countries may decide notto take any significant actions to impact economicgrowth in order to reduce their CO2 emissions. How-ever, in the case of Indonesia and the People’s Republicof China, where CO2 emissions and economic growthwere found to have a positive relationship, an increase inGDP will not reduce CO2 emissions.As for the Granger causality between energy consump-

tion and economic growth, energy consumptionGranger-cause economic growth in Brazil, India,Indonesia, Mexico and Turkey is consistent with the factthat these developing countries are at their industrializedstage. It also supports the “energy led-growth hypoth-esis” which predicts that, more often than not, energyprotection policies are bound to stir economic growth.There was also an indication of economic growthGranger-cause energy consumption in Brazil, India,Indonesia, Mexico, and the People’s Republic of China.This suggests that if economic growth increases energyconsumption, the externality created out of energy usewill ultimately reverse into economic growth. Under thiscircumstance, conservation policies are urgentlyrequired.

Table 7 Long-run causality test

Country LGDP LEC LCO2

F statistics (p value) (sign) F statistics (p value) (sign) F statistics (p value) (sign)

Brazil LCO2 2.136213/[0.1536] (–) 2.460071 / [0.1266] (+) – –

India LEC 5.297746**/[0.0328] (+) – – 1.561488/[0.2266] (+)

Russia LEC 0.352065/[0.5640] (–) – – 0.699432/[0.4193] (+)

Note: Asterisk ** indicates the 5% significance levelSquare brackets indicate the characterization factor of the p value. No cointegration and its causality test involve only lag variables

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 13 of 17

The bootstrap ARDL bound test enabled an under-standing of the long-term relationship among the 3Evariables and used this relationship to predict changes inCO2 emissions and energy consumption in the E7 coun-tries over time. Similarly, the 3E relationship must be

considered in light of the particular conditions of eachcountry, because each country presents a unique com-posite of social, legal, economic, political, and techno-logical characteristics. The empirical results providepolicymakers with a better understanding of the linkages

LGDP

LEC

P.R. China

Brazil Indonesia India

Russia

Mexico

Turkey

LGDP

LEC LCO2

LGDP

LEC

LGDP

LEC

LGDP

LEC

LGDP

LEC LCO2

LGDP

LECLCO2 LCO

2LCO

2

LCO2

LCO2

Fig. 2 3E Granger causality relationships for E7 countries

Table 8 Results for Granger causality test based on ARDL model

Country LGDP equation LEC equation LCO2 equation

F statistics (p value) F statistics (p value) F statistics (p value)

Brazil LGDP – – 2.405635*/[0.0800] (+) 3.767159**/[0.0177] (–)

LEC 8.200355***/[0.0005] (+) – – 8.665529***/[0.0004] (–)

LCO2 8.047049***/[0.0017] (–) 9.973761***/[0.0005] (+) – –

India LGDP – – 10.01860***/[0.0002] (+) 2.494419*/[0.0842] (–)

LEC 38.67256***/[0.0000] (+) – – 59.24849***/[0.0000] (+)

LCO2 3.277671**/[0.0338] (+) 4.296970**/[0.0126] (–) – –

Indonesia LGDP – – 2.728564*/[0.0827] (–) 0.485877/[0.6202] (no)

LEC 2.471449*/[0.0768] (–) – – 1.661532/[0.2009] (no)

LCO2 3.369837**/[0.0303] (+) 2.825815*/[0.0769] (+) – –

Mexico LGDP – – 2.641577*/[0.0607] (+) 6.732778***/[0.0012] (–)

LEC 29.22243***/[0.0000] (+) – – 4.472803***/[0.0090] (+)

LCO2 4.520791***/[0.0084] (–) 5.014765***/[0.0053] (–) – –

People’s Republic of China LGDP – – 5.734735**/[0.0227] (–) 10.60359***/[0.0027] (+)

LEC 0.109901/[0.9534] (no) – – 7.015114***/[0.0016] (+)

LCO2 6.805757***/[0.0019] (–) 4.463740**/[0.0130] (+) – –

Russia LGDP – – 0.575145/[0.5843] (no) 0.828062/[0.4711] (no)

LEC 0.091798/[0.7671] (no) – – 30.43173***/[0.0000] (+)

LCO2 6.120302**/[0.0293] (–) 26.29907***/[0.0000] (+) – –

Turkey LGDP – – 1.556601/[0.2136] (no) 1.348327/[0.2815] (no)

LEC 12.47108***/[0.0001] (+) – – 12.64742***/[0.0001] (+)

LCO2 30.21155***/[0.0000] (+) 9.017897***/[0.0001] (+) – –

Note: Asterisks ***, ** and * indicate the 1%, 5% and 10% significance levels, respectivelySquare brackets indicate the characterization factor of the p value. No cointegration and its causality test involve only lag variables

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 14 of 17

between economic growth, energy consumption, andCO2 emissions to fine-tune their energy policies.Policymakers should aim to improve energy efficiency

based on a robust infrastructure that will enable busi-nesses and households to use fossil fuels to generateclean energy. Their countries need to upgrade theirhigh-cost/low-efficient energy technologies used formanufacturing goods to more efficient and less energy-dependent ones, in order to create a better environmentand to promote societal changes. Specifically, thesecountries must invest in renewable energy and the pro-motion of technology transfers, to facilitate the gener-ation of renewable energy.

ConclusionsIn recent years, the economic growth–energy consump-tion–environment trinity has become a key issue forscholars and policymakers. Increased demand for energyand the subsequent CO2 pollution make both developedand developing countries vulnerable to the effects of glo-bal warming and climate change. This compels govern-ments to create policies which aim to reduce CO2

emissions while concurrently creating new knowledgeand expertise to supply the energy that is required forgrowth and development. However, while the worldwideeffort is to achieve more efficient use of energy, E7countries face a more substantial challenge. They havethe task of achieving economic growth while simultan-eously improving the efficiency of their energy consump-tion, reducing environmental pollution, and promotingCO2 emission-reduction strategies.The political drive to take action against global warm-

ing is a sine qua non condition for both developed anddeveloping countries. This motivates the creation andenactment of policies to produce environmentally sus-tainable energy and regulation of CO2 emissions result-ing from the combustion of fossil fuels. It is, at the sametime, the most questioned climate-friendly energy policy.This is because some countries see CO2 emissions as anirreversible threat to the planet while others, with poorenvironmental records, see it as an unfounded exagger-ation lacking in any scientific evidence. The result of thisconflict of attitude is an international diplomatic dead-lock between proactive and lukewarm nations regardingthe amelioration of CO2 emissions and promotion ofsources of renewable energy. Such an impasse is com-pounded in the E7 countries by the sheer complexity oftheir environmental problems and pollution issues, therelatively high cost of carbon sequestration technologies,and the lack of mandatory enforceability. Consequently,unless the E7 nations secure an efficient infrastructurefor energy conversion to adequately deliver variousforms of clean energy at affordable prices, they will allfail to meet one of the most fundamental sustainable

development goals advanced by the United Nations, thatof reaching net zero CO2 emissions by 2030.Prior to this study, no research using the bootstrap

ARDL bound test had been carried out on the E7 coun-tries and only limited research existed on the directionof the causal relationship between economic growth, en-ergy consumption, and CO2 emissions in general. Thispaper has responded to the call for research that enablesgovernments to appropriately capture actual riskthrough stress testing using novel research methodolo-gies. It has filled the gap in current literature by revisit-ing the cointegration and causality relations of 3E in theE7 countries with a distinctive variation of the bootstrapARDL bound model. The results show there is a signifi-cant long-run cointegration among economic growth,energy consumption, and CO2 emissions, while evidencefor cointegration exists for Brazil, India, and Russia.This study confirms the presence of Granger causality

from energy consumption to CO2 emissions. The globalwarming consequences of CO2 emissions have made itnecessary for developed and developing countries to re-duce fossil fuel consumption and promote the use ofclean energy. The one-way and two-way causalities be-tween economic growth and CO2 emissions, energy con-sumption, and CO2 emissions, and energy consumptionand economic growth, also indicate that energy con-sumption is still growing in the E7 countries. Accom-plishing significant global reductions in CO2 emissionsrequires both the tradeoff of slower economic growthand the achievement of coordinated sustainable environ-mental development. Standardization of research ap-proaches would permit balancing out environmentalrisks and steady growth, while allowing each governmentto formulate specific energy development, CO2 reduc-tion, and economic policies that encourage the use of re-newable sources of energy while reflecting its owncountry’s unique circumstances.The results underscore the intertwined relationship

between economic growth–energy consumption–envir-onment in the E7 countries. Future research should lookat identifying and quantifying the short- and long-termbenefits and costs of CO2 reduction and mitigation strat-egies. This would permit policymakers to design the in-stitutional framework for unified national energystandards, along with an assortment of subsidies andtaxes and building and transportation codes, beforemaking decisions on the alternatives available for dis-rupting the frequency and intensity of natural disastersassociated with CO2 emissions. These alternativesshould allow the E7 countries to move away from fossilfuels and focus on alternative technologies, includingsolar and wind as well as the development of low-carbonpower plants based on gas and steam turbines. This re-search is not exempt from some limitations. A case in

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 15 of 17

point is the data collection which could have included amuch longer span time for each E7 country to capturemore precisely their 3E relationships. Further researchalso needs to cast the net to a wider spectrum of well-known global leaders of CO2 emissions to ensure theseresults hold in nations with unique social, legal, eco-nomic, political, and technological characteristics. Thereis still time for these governments to become moreforceful in their efforts to act in order to slow down thepace with which our climate is changing around theworld.

AbbreviationsARDL: Autoregressive distributed lagged (ARDL); E7 countries: Emerging 7countries (Brazil, India, Indonesia, Mexico, People’s Republic of China, Russia,and Turkey); GDP: Gross domestic product; 3E: Economic growth, energyconsumption, and CO2 emissions; CO2: Carbon dioxide; BP: British Petroleum;SDGs: Sustainable Development Goals; IEA: International Energy Agency;PWC: Price, Waterhouse, and Coopers; G7: Group of 7; EKC: EnvironmentalKuznets Curve; BRIC: Brazil, Russia, India, and People’s Republic of China;EC: Energy consumption; FD: Financial development; VAR: VectorAutoregressive Model; VECM: Vector Error Correction Model; Kgoe: Kilogramof oil equivalent; mt: Metric tons; ADF: Augmented Dickey–Fuller; PP: Phillipsand Perron; KPSS: Kwiatkowski–Phillips–Schmidt–Shin; LGDP: Logarithm ofper-capita GDP; LEC: Logarithm of per-capita primary energy consumption;LCO2: Logarithm of per-capita CO2 emissions; LS: Lee and Strazicich; Z-A: Zivot and Andrew; AIC: Akaiki Information Criterion

AcknowledgementsNone.

Authors’ contributionsJaime Ortiz supervised the research from the beginning to the end. TengTong conceived the study and drafted the manuscript. Chuanhua Xucollected the data. Fangjhy Li estimated the econometric model. All authorsmade suggestions throughout the entire manuscript. The authors read andapproved the final manuscript.

FundingThis study was supported by the Hubei Provincial Department of EducationHumanities and Social Sciences project (funding no. 16Q207), the ResearchCenter of Hubei Financial Development and Financial Security, and thePeople’s Republic of China Scholarship Council (CSC 201808420340).

Availability of data and materialsAll data used in this study are publicly available online. The information onthe sources of data is provided in the references.

Ethics approval and consent to participateNot applicable

Consent for publicationNot applicable

Competing interestsThe authors declare that they have no competing interests.

Author details1School of Finance, Hubei University of Economics, Wuhan 430205, People’sRepublic of China. 2College of Technology, University of Houston, Isabel C.Cameron Bldg. Suite. 227 D, 4235 Cullen Blvd., Houston, TX 77204, USA.

Received: 20 December 2019 Accepted: 30 March 2020

References1. Cheng BS, Lai TW (1997) An investigation of co-integration and causality

between energy consumption and economic activity in Taiwan. EnergyEcon 19(4):435–444

2. Ghosh S (2002) Electricity consumption and economic growth in India.Energy Policy 30:125–129

3. Kraft J, Kraft A (1978) On the relationship between energy and GNP. JEnergy Dev.:401–403

4. Lee CC (2005) Energy consumption and GDP in developing countries: acointegrated panel analysis. Energy Econ. 27:415–427. https://doi.org/10.1016/j.eneco.2005.03.003

5. Jobert T, Karanfil F (2007) Sectoral energy consumption by source andeconomic growth in Turkey. Energy Policy 35:5447–5456. https://doi.org/10.1016/j.enpol.2007.05.008

6. Apergis N, Payne JE (2009) CO2 emissions, energy usage, and output inCentral America. Energy Policy 37:3282–3286. https://doi.org/10.1016/j.enpol.2009.03.048

7. Naser H (2015) Analyzing the long-run relationship among the oil market,nuclear energy consumption, and economic growth: an evidence fromemerging economies. Energy 89:421–434. https://doi.org/10.1016/j.energy.2015.05.115

8. Chen PY, Chen ST, Hsu CS, Chen CC (2016) Modeling the globalrelationships among economic growth, energy consumption, and CO2

emissions. Renew Sustain Energy Rev. 65:420–431. https://doi.org/10.1016/j.rser.2016.06.074

9. Sadorsky P (2009) Renewable energy consumption, CO2 emissions and oilprices in the G7 countries. Energy Econ. 31:456–462. https://doi.org/10.1016/j.eneco.2008.12.010

10. Wang SS, Zhou DQ, Zhou P, Wang QW (2011) CO2 emissions, energyconsumption and economic growth in China: a panel data analysis. EnergyPolicy 39:4870–4875. https://doi.org/10.1016/j.enpol.2011.06.032

11. Raza SA, Shahbaz M, Nguyen DK (2015) Energy conservation policies,growth, and trade performance: evidence of feedback hypothesis inPakistan. Energy Policy 80:1–10. https://doi.org/10.1016/j.enpol.2015.01.011

12. Grippa P, Schmittmann J, Suntheim F (2019) Climate change and financialrisk. Finance Dev 56(4):26–29

13. United Nations, "Transforming our world: the 2030 agenda for sustainabledevelopment". – Sustainable development knowledge platform. Retrieved23 August 2015.

14. Cui L, Huang Y (2018) Exploring the schemes for green climate fundfinancing: international lessons. World Dev. 101:173–187. https://doi.org/10.1016/j.worlddev.2017.08.009

15. Doğan B (2018) The role of economic growth and energy consumption onCO2 emissions in E7 countries XXV, 231–246

16. International Energy Agency (IEA), https://www.iea.org/geco/.17. PWC, The Long view, How will the global economic order change by 2050?

(https://www.pwc.com.au/government/pwc-the-world-in-2050-full-report-feb-2017.pdf)

18. Hawksworth, John, Cookson Gordon.2006. "The world in 2050" .PricewaterhouseCoopers.

19. Park G (2016) Integral operational leadership: a relationally intelligentapproach to sustained performance in the twenty-first century. RoutledgeISBN 9781317070863

20. BP Statistical Review of World Energy Reports, 2018. (https://www.bp.com/en/global/corporate/energy-economics/statistical-review-of-world-energy.html)

21. Pao H-T, Chen C-C (2019) Decoupling strategies: CO2 emissions, energyresources, and economic growth in the Group of Twenty. J Clean Prod. 206:907–919. https://doi.org/10.1016/j.jclepro.2018.09.190

22. Yang Z, Zhao Y (2014) Energy consumption, carbon emissions, andeconomic growth in India: evidence from directed acyclic graphs. EconModel. 38. https://doi.org/10.1016/j.econmod.2014.01.030

23. Ozturk I, Acaravci A (2010) CO2 emissions, energy consumption andeconomic growth in Turkey. Renew Sustain Energy Rev. 14:3220–3225.https://doi.org/10.1016/j.rser.2010.07.005

24. Asongu S, El Montasser G, Toumi H (2016) Testing the relationshipsbetween energy consumption, CO2 emissions, and economic growth in 24

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 16 of 17

African countries: a panel ARDL approach. Environ Sci Pollut Res. 23:6563–6573 https://doi.org/10.1007/s11356-015-5883-7

25. Ahmad N, Du L (2017) Effects of energy production and CO2 emissions oneconomic growth in Iran: ARDL approach. Energy 123:521–537 https://doi.org/10.1016/j.energy.2017.01.144

26. Pesaran MH, Shin Y, Smith RJ (2001) Bounds testing approaches to theanalysis of level relationships. J Appl Econ. 16:289–326 https://doi.org/10.1002/jae.616

27. McNown R, Sam CY, Goh SK (2018) Bootstrapping the autoregressivedistributed lag test for cointegration. Appl Econ. 50(13):1509e1521

28. Grossman, G.M., Krueger, A.B., 1995. Economic growth and the environment.Quarterly Journal of Economics, 60(2), 353–375.Published by : The MIT PressStable URL : http://www.jstor.org/stable/2118443. Q. J. Econ. 110, 353–377.

29. Dinda S (2004) Environmental Kuznets curve hypothesis: a survey. Ecol Econ.49:431–455 https://doi.org/10.1016/j.ecoLECon.2004.02.011

30. Soytas U, Sari R, Ewing BT (2007) Energy consumption, income, and carbonemissions in the United States. Ecol. Econ. 62:482–489 https://doi.org/10.1016/j.ecolecon.2006.07.009

31. Iwata H, Okada K, Samreth S (2010) Empirical study on the environmentalKuznets curve for CO2 in France: the role of nuclear energy. Energy Policy38:4057–4063 https://doi.org/10.1016/j.enpol.2010.03.031

32. Dietz T, Rosa EA (1994) Rethinking the environmental impacts ofpopulation, affluence and technology. Human Ecology Review 1:277–300

33. Özokcu S, Özdemir Ö (2017) Economic growth, energy, and environmentalKuznets curve. Renew Sustain Energy Rev. 72:639–647 https://doi.org/10.1016/j.rser.2017.01.059

34. Friedl B, Getzner M (2003) Determinants of CO2 emissions in a small openeconomy. Ecol Econ 45:133–148. https://doi.org/10.1016/S0921-8009(03)00008-9

35. Holtz-Eakin D, Selden TM (1995) CO2 emissions and economic growth. JPublic Econ. 57:85–101

36. Agras J, Chapman D (1999) A dynamic approach to the environmentalKuznets curve hypothesis. Ecol Econ 28:267–277. https://doi.org/10.1016/S0921-8009(98)00040-8

37. He J, Richard P (2010) Environmental Kuznets curve for CO2 in Canada. EcolEcon. 69:1083–1093 https://doi.org/10.1016/j.ecoLECon.2009.11.030

38. Richmond AK, Kaufmann RK (2006) Is there a turning point in therelationship between income and energy use and/or carbon emissions?Ecol. Econ. 56:176–189 https://doi.org/10.1016/j.ecoLECon.2005.01.011

39. Stern, D.I., 1993. Energy and economic growth in the USA. Energy Econ. 15,137–150. https://doi.org/10.1016/0140-9883 (93)90033-N

40. Salahuddin M, Gow J, Ozturk I, 2015. Is the long-run relationship betweeneconomic growth, electricity consumption, carbon dioxide emissions andfinancial development in Gulf Cooperation Council Countries robust?Renew Sustain Energy Rev. 51, 317–326. https://doi.org/10.1016/j.rser.2015.06.005

41. Omri A, Nguyen DK, Rault C (2014) Causal interactions between CO2

emissions, FDI, and economic growth: evidence from dynamicsimultaneous-equation models. Econ Model. 42:382–389 https://doi.org/10.1016/j.econmod.2014.07.026

42. Ozturk I (2010) A literature survey on energy-growth nexus. Energy Policy38:340–349 https://doi.org/10.1016/j.enpol.2009.09.024

43. Damette O, Seghir M (2013) Energy as a driver of growth in oil exportingcountries? Energy Econ. 37:193–199 https://doi.org/10.1016/j.eneco.2012.12.011

44. Ghali KH, El-Sakka MIT (2004) Energy use and output growth in Canada: amultivariate cointegration analysis. Energy Econ. 26:225–238 https://doi.org/10.1016/S0140-9883

45. Baranzini A, Weber S, Bareit M, Mathys NA (2013) The causal relationshipbetween energy use and economic growth in Switzerland. Energy Econ. 36:464–470 https://doi.org/10.1016/j.eneco.2012.09.015

46. Jamil F, Ahmad E (2010) The relationship between electricity consumption,electricity prices and GDP in Pakistan. Energy Policy 38:6016–6025 https://doi.org/10.1016/j.enpol.2010.05.057

47. Belloumi M (2009) Energy consumption and GDP in Tunisia: cointegrationand causality analysis. Energy Policy 37:2745–2753 https://doi.org/10.1016/j.enpol.2009.03.027

48. Lee CC, Chang CP, Chen PF (2008) Energy-income causality in OECDcountries revisited: The key role of capital stock. Energy Econ. 30:2359–2373https://doi.org/10.1016/j.eneco.2008.01.005

49. Tang CF, Tan BW (2014) The linkages among energy consumption,economic growth, relative price, foreign direct investment, and financialdevelopment in Malaysia. Qual Quant. 48:781–797 https://doi.org/10.1007/s11135-012-9802-4

50. Pao HT, Tsai CM (2010) CO2 emissions, energy consumption and economicgrowth in BRIC countries. Energy Policy 38:7850–7860 https://doi.org/10.1016/j.enpol.2010.08.045

51. Wang K, Zhu B, Wang P, Wei YM (2016) Examining the links amongeconomic growth, energy consumption, and CO2emission with linear andnonlinear causality tests. Nat Hazards 81:1147–1159 https://doi.org/10.1007/s11069-015-2124-9

52. Magazzino C (2016) Economic growth, CO2 emissions and energy use inthe south caucasus and Turkey: a PVAR analyses. Int Energy J. 16:153–162

53. Magazzino C (2017) The relationship among economic growth, CO2

emissions, and energy use in the APEC countries: a panel VAR approach.Environ Syst Decis. 37:353–366 https://doi.org/10.1007/s10669-017-9626-9

54. Mirza FM, Kanwal A (2017) Energy consumption, carbon emissions andeconomic growth in Pakistan: dynamic causality analysis. Renew SustainEnergy Rev. 72:1233–1240 https://doi.org/10.1016/j.rser.2016.10.081

55. Jalil A, Mahmud SF (2009) Environment Kuznets curve for CO2 emissions: acointegration analysis for China. Energy Policy 37:5167–5172 https://doi.org/10.1016/j.enpol.2009.07.044

56. Magazzino C (2016) The relationship among real gross domestic product,CO2 emissions, and energy use in South Caucasus and Turkey. Int J EnergyEcon Policy 6(4):672–683

57. Alam MM, Murad MW, Noman AHM, Ozturk I (2016) Relationships amongcarbon emissions, economic growth, energy consumption and populationgrowth: testing environmental Kuznets curve hypothesis for Brazil, China,India and Indonesia. Ecol Indic. 70:466–479 https://doi.org/10.1016/j.ecolind.2016.06.043

58. Goh SK, Yong JY, Lau CC, Tang TC (2017) Bootstrap ARDL on energy-growthrelationship for 22 OECD countries. Appl Econ Lett. 24:1464–1467 https://doi.org/10.1080/13504851.2017.1284980

59. Lin FL, Inglesi-Lotz R, Chang T (2018) Revisit coal consumption, CO2

emissions and economic growth nexus in China and India using a newlydeveloped bootstrap ARDL bound test. Energy Explor Exploit. 36:450–463https://doi.org/10.1177/0144598717741031

60. Phillips P, Perron P (1988) Testing for a unit root in time series regression.Biometrika 75(2):335–346

61. Elliott G, Rothenberg TJ, Stock JH (1996) Efficient tests for an autoregressiveunit root. Econometrica 64(4):813–836

62. Kwiatkowski D, Phillips PCB, Schmidt P, Shin Y (1992) Testing the nullhypothesis of stationarity against the alternative of a unit root. How sureare we that economic time series have a unit root? J Econom. 54:159–178https://doi.org/10.1016/0304-4076(92)90104-Y

63. Zivot E, Andrews D (1992) Further evidence on the great crash, the oil priceshock, and the unit-root hypothesis. J Business Econ Stat 10(3):251–270

64. Lee J, Strazicich MC (2003) Minimum Lagrange multiplier unit root test withtwo structural breaks. Rev Econ Stat 85(4):1082–1089 https://doi.org/10.1162/003465303772815961

65. Bai J, Perron P (2003) Computation and analysis of multiple structuralchange models. J Appl Econom. 18:1–22 https://doi.org/10.1002/jae.659

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Tong et al. Energy, Sustainability and Society (2020) 10:20 Page 17 of 17